| shtylman | cjwatson: what were those main keyboard layouts you told me about at UDS? the ones I should focus on? ... this is why I want replay on life :) | 03:52 |
|---|---|---|
| shtylman | and is there a wiki page or something with some pretty pictures? | 04:02 |
| CIA-15 | ubiquity: superm1 * r3614 ubiquity/ (debian/changelog ubiquity/components/ubi-intro.py): | 09:09 |
| CIA-15 | ubiquity: Correct a variable that didn't get changed from pychecker | 09:09 |
| CIA-15 | ubiquity: cleanup in the intro plugin. | 09:09 |
| === cjwatson_ is now known as cjwatson | ||
| cjwatson | shtylman: Brazilian and Japanese, since they're 106-key layouts | 09:41 |
| ev | argh. Just realized a nasty corner case of copying the keyring and relevant portion of gconf (/system/networking) to the target system. What if the user didn't elect to format /home or /? | 13:07 |
| ev | hrm, I guess I could just check for existence of both and do nothing if they're already there | 13:08 |
| === robbiew-afk is now known as robbiew | ||
| cr3 | if I preseed repositories with local0/repository, will the latest packages be retrieved from that location during the installation or will that only result in the repositories being added to the sources.list file after the installation has completed? | 13:50 |
| shtylman | cjwatson: what about british keyboards? | 13:50 |
| shtylman | anything different in key number there? or still 104 key? | 13:50 |
| cjwatson | shtylman: they fall into the giant category of "everything that isn't us, br, or jp" | 13:50 |
| cjwatson | *everything* else is 105-key, at least to an excellent first approximation | 13:51 |
| cjwatson | us is the only 104-key layout I know of | 13:51 |
| shtylman | ahh ok | 13:51 |
| shtylman | well... that gives me a starting point :) | 13:51 |
| cjwatson | cr3: should be the former, at least for netboot installations. CD installations are a bit different because in general we want to prefer stuff from the CD over a giant download | 13:51 |
| cr3 | cjwatson: I'm thinking that if local0/repository is defined explicitly, then the user gets what he's asking for. however, for the CD, if that means that the default repository is set to the CD itself and local0/repository to archive.u.c, then I can appreciate that use case | 13:53 |
| cr3 | (where archive.u.c is just shorthand for whatever magic the installer does to determine the closest mirror) | 13:54 |
| cjwatson | um | 13:55 |
| cjwatson | archive.u.c is handled separately. apt-setup/local0/repository is strictly for local overrides only | 13:55 |
| cjwatson | you might be right that it might make sense to handle local* differently on CD installs, although unfortunately it's actually kind of hard to unpick that by the time we notice ... | 13:56 |
| cjwatson | feel free to file a bug on pkgsel to improve that | 13:56 |
| CIA-15 | base-installer: cjwatson * r389 ubuntu/ (debian/changelog library.sh): | 15:03 |
| CIA-15 | base-installer: Explicitly ignore Recommends while installing the kernel. We don't want | 15:03 |
| CIA-15 | base-installer: to install bootloaders at this point. | 15:03 |
| CIA-15 | base-installer: cjwatson * r390 ubuntu/debian/changelog: releasing version 1.103ubuntu2 | 15:05 |
| === rgreening_ is now known as rgreening | ||
| CIA-15 | cdrom-detect: cjwatson * r455 ubuntu/debian/ (cdrom-detect.postinst changelog): | 17:26 |
| CIA-15 | cdrom-detect: Copy /cdrom/.disk/info to /var/log/media-info, in order that we get | 17:26 |
| CIA-15 | cdrom-detect: /var/log/installer/media-info on the installed system; doing this in | 17:26 |
| CIA-15 | cdrom-detect: save-logs is too late because /cdrom is already unmounted by that point | 17:26 |
| CIA-15 | cdrom-detect: (LP: #364649). | 17:26 |
| CIA-15 | cdrom-detect: cjwatson * r456 ubuntu/debian/changelog: releasing version 1.32ubuntu2 | 17:29 |
| CIA-15 | installation-report: cjwatson * r71 ubuntu/ (debian/changelog finish-install.d/94save-logs): | 17:30 |
| CIA-15 | installation-report: Revert 2.39ubuntu2, which didn't work as /cdrom is unmounted by this | 17:30 |
| CIA-15 | installation-report: point. This is now handled in cdrom-detect instead. | 17:30 |
| CIA-15 | installation-report: cjwatson * r72 ubuntu/debian/changelog: releasing version 2.39ubuntu3 | 17:31 |
| bondiblueos9 | hey | 18:35 |
| bondiblueos9 | I'd like to install ubuntu on my powerpc mac, but my cd drive is broken | 18:36 |
| === robbiew is now known as robbiew-afk | ||
| corp186 | can someone point me to documentation on how to maintain a local repo and add new udebs to it for development purposes? | 19:52 |
| corp186 | I think I figured out how to generate a new Packages file using dpkg-scanpackages | 20:13 |
| corp186 | now I think I'm having signing issues: | 20:13 |
| corp186 | net-retriever: gpgv: BAD signature from "Ubuntu Archive Automatic Signing Key <ftpmaster@ubuntu.com>" | 20:13 |
| corp186 | anna[2013]: cat: can't open '/tmp/net-retriever-2017-deduplicate/*'; No such file or directory | 20:14 |
| corp186 | anna[2013]: WARNING **: bad d-i Packages file | 20:15 |
| cjwatson | delete Release.gpg, you can't replicate it | 20:15 |
| cjwatson | there's a documented preseed to disable authentication | 20:16 |
| corp186 | so I need a preseed on top of deleting Release.gpg? | 20:16 |
| cjwatson | 'd-i debian-installer/allow_unauthenticated string true' | 20:16 |
| cjwatson | it's also possible to feed in a key using apt-setup/local0/key, but that's more work | 20:16 |
| cjwatson | (both preseeds are documented in Appendix B to the installation guide) | 20:17 |
| corp186 | cjwatson: network preseeding occurs early enough to be effective? | 20:19 |
| corp186 | or do I need to do initrd preseeding, since I'm doing a netboot setup | 20:19 |
| corp186 | ok, so the preseed worked for authentication, but now I see that I need to regenerate the Release file | 20:31 |
| corp186 | is there some utility to do that? | 20:31 |
| cjwatson | apt-ftparchive | 20:32 |
| cjwatson | (supersedes dpkg-scanpackages) | 20:32 |
| corp186 | cjwatson: thanks | 20:35 |
| corp186 | I think I've finally gotten everything I need to start some dev work | 20:35 |
| corp186 | cjwatson: one architectural question: when I need to bypass a later installation task (like user-setup) I figured I would touch a file somewhere and check for it in the user-setup postinst script before continuing | 20:36 |
| corp186 | does this seem reasonable, and if so where would you put the "stamp" file | 20:37 |
| cjwatson | you're bypassing things conditionally? | 20:39 |
| cjwatson | (where: /var/run/) | 20:39 |
| corp186 | cjwatson: yes | 20:39 |
| cjwatson | seems ok, though if you have to modify user-setup.postinst anyway you could just modify it to do what you want | 20:40 |
| cjwatson | I wouldn't put something with stamp files like that into Ubuntu but it's OK for a derivative | 20:40 |
| corp186 | so here's my idea in a nutshell: proper backup and restore | 20:40 |
| corp186 | have a utility that only backs up the packages that are currently installed along with /etc, /home/ /usr/local etc. | 20:41 |
| corp186 | then at install time (or restore time as it may be), you can restore from the backup by: | 20:42 |
| corp186 | 1. copying old /etc/passed and /etc/group into /target | 20:42 |
| corp186 | 2. install old /etc/apt/sources* into /target | 20:42 |
| corp186 | 3. install all packages from backup list (exact same packages versions, not the latest versions) | 20:43 |
| corp186 | 4. merge hw specific /etc config files (i.e. don't overwrite things like xorg.conf or module blacklists in case the hw platform has changed) | 20:44 |
| corp186 | 5. copy over the rest of the backed up files | 20:44 |
| corp186 | but this necessitates skipping over user-setup | 20:44 |
| cjwatson | right. sounds worthwhile, though I imagine there are lots of scary corner cases and so I'm more than happy to let you prototype it ... :-) | 20:44 |
| cjwatson | (e.g. sometimes the package list varies slightly depending on the hardwarE) | 20:45 |
| cjwatson | s/E/e/ | 20:45 |
| corp186 | cjwatson: yeah, lots of corner cases, but might as well try it and get something that works for many/most people | 20:45 |
| cjwatson | a stamp file is the easiest approach; maybe later it'll be worth figuring out something more elegant | 20:46 |
| corp186 | cjwatson: I'm getting an error now on debootstrap: | 20:46 |
| corp186 | E: NOSCRIPT | 20:46 |
| corp186 | EA: /usr/share/debootstrap/scripts/unknown | 20:47 |
| corp186 | EF: No such script: %s | 20:47 |
| corp186 | any idea what that means before I do more in depth debugging? | 20:47 |
| cjwatson | are you using a modified CD image? | 20:47 |
| corp186 | I'm doing netboot | 20:47 |
| corp186 | I couldn't figure out how to make a full cd | 20:47 |
| corp186 | and netboot seems to work well enough so far | 20:48 |
| cjwatson | probably means you forgot to fill in the Suite and/or Codename fields in your Release file | 20:48 |
| cjwatson | compare http://archive.ubuntu.com/ubuntu/dists/lucid/Release | 20:48 |
| corp186 | actually, I'm still working off of karmic, cause I don't want to deal with lucid breaking things as I prototype | 20:49 |
| corp186 | it had been working fine until I regnereated Packages and Release | 20:49 |
| cjwatson | lucid vs. karmic wasn't significant in my comment | 20:49 |
| cjwatson | check your regenerated Release file :-) | 20:49 |
| cjwatson | I bet it says "unknown" in it somewhere | 20:50 |
| corp186 | I actually didn't "regen" the Release file; I had already manually updated it with new md5sums and sizes before I saw your comment about apt-ftparchive | 20:50 |
| corp186 | wait... | 20:51 |
| corp186 | you're right | 20:51 |
| corp186 | codename is unknown somehow | 20:51 |
| corp186 | oh.... I had played around with apt-move | 20:51 |
| corp186 | and it mucked it up | 20:51 |
| corp186 | base system is installing right now | 20:56 |
| corp186 | thanks :) | 20:56 |
| soren | Are the no cases anymore where the installer chooses lilo over grub? | 21:24 |
| cjwatson | not TTBOMK | 21:25 |
| soren | All the LVM or XFS special cases are gone? Neat. | 21:25 |
| cjwatson | although the manual option is still there | 21:25 |
| cjwatson | (grub-installer/skip=true) | 21:25 |
| soren | I didn't realise. I haven't done an install for my own needs for a while. | 21:25 |
| cjwatson | possible compromise here: take lilo off the CDs, but leave it in supported | 21:26 |
| cjwatson | that way if people come to us saying "argh, I needed lilo" we can (a) point them at netboot (b) find out why | 21:26 |
| cjwatson | it will help with (b) because they'll have to make a conscious change | 21:26 |
| cjwatson | but won't leave them completely out in the cold | 21:27 |
| cjwatson | grub2 covers LVM, and we fixed the XFS problems a while back | 21:27 |
| soren | Right. You did see mathiaz' e-mail to ubuntu-devel earlier today, right? lilo is suggested for demotion to universe. | 21:27 |
| cjwatson | not yet though I was aware of the general proposal from a UDS discussion | 21:28 |
| cjwatson | I'll follow up with my counterproposal | 21:28 |
| soren | Wicked. Just wanted to make sure it was on your radar. | 21:28 |
| === robbiew-afk is now known as robbiew | ||
| corp186 | cjwatson: you mentioned that apt-ftparchive can generate the Release file in the repo, but I haven't figured out how to do that | 22:37 |
| corp186 | can you explain how you do it? | 22:37 |
| cjwatson | I don't do this very often, but it's 'apt-ftparchive release' or '... generate'. However it's very late for me, I'll have to refer you to the docs or the web | 22:38 |
| corp186 | hmm, the man page didn't mention any release command | 22:38 |
| corp186 | ugh... | 22:38 |
| corp186 | I was looking at an apparently out of date man page on line | 22:38 |
| corp186 | thanks | 22:39 |
| === robbiew is now known as robbiew-afk | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!